Contents

1 Introduction

To evaluate the aneuploidy and prevalence of clonal or quasiclonal tumors, we developed a novel tool to characterize the mosaic tumor genome on the basis of one major assumption: the genome of a heterogeneous multi-cell tumor biopsy can be sliced into a chain of segments that are characterized by homogeneous somatic copy number alternations (SCNAs) and B allele frequencies (BAFs). The model, termed BubbleTree, utilizes both SCNAs and the interconnected BAFs as markers of tumor clones to extract tumor clonality estimates. BubbleTree is an intuitive and powerful approach to jointly identify ASCN, tumor purity and (sub)clonality, which aims to improve upon current methods to characterize the tumor karyotypes and ultimately better inform cancer diagnosis, prognosis and treatment decisions.

1.1 Quickstart to Using BubbleTree

To perform a BubbleTree analysis, data pertaining to the position and B allele frequency of heterozygous snps in the tumor sample and segmented copy number information including the position, number of markers/segment and log2 copy number ratio between tumor and normal samples must first be obtained.

1.2 Preparing Data for BubbleTree

BubbleTree was developed using both whole exome sequencing (WES) and whole genome sequening (WGS) NGS data from paired tumor/normal biopsies, but this model should also be applicable to array comparative genomic hybridization (aCGH) and single nucleotide polymorphism (SNP) array data.

There are many methods for generating and processing sequencing data in preparation for BubbleTree analysis. In the following section we provide example workflows starting from WES NGS which can be adapted as needed to alternate inputs.

1.3 Preparing Sequence Variation Data

The primary BubbleTree requirement for sequence variant information is a GRanges object containing the B alelle frequencies and genomic positions of variants known to be heterozygous in the paired normal sample.

Mapped reads from tumor and normal tissue can be processed with mutation caller software such as VARSCAN or MUTECT. In this example, we use a hypothetical vcf file output which contains mutation calls from both normal and tumor samples.

The B-allele frequency data is extracted using the Bioconductor package VariantAnnotation and converted from string to numeric format.

Example data in the desired format is provided as part of this package as GRanges objects and can be loaded as follows.

library(BubbleTree)

allCall.lst is pre-calculated CNV data. allRBD.lst is simply the RBD data from below.

load(system.file("data", "allCall.lst.RData", package="BubbleTree"))
head(allCall.lst[[1]]@rbd)
##   seqnames    start       end     width strand seg.id num.mark    lrr
## 1    chr10    93890  38769716  38675827      *    806    31699 0.1413
## 2    chr10 38877329 135523936  96646608      *    808    74425 0.1415
## 3    chr11   133952 134946370 134812419      *    812   102934 0.1412
## 4    chr12    60000 133841793 133781794      *    813   103392 0.1413
## 5    chr13 19020000 115109861  96089862      *    814    76080 0.1419
## 6    chr14 20191636 107288640  87097005      *    823    68709 0.1425
##       kurtosis        hds     hds.sd het.cnt seg.size
## 1 -0.093044958 0.01851852 0.06051370   10400 1.487216
## 2 -0.048701274 0.01851852 0.06046402   25414 3.491784
## 3 -0.018840021 0.01851852 0.06052843   36183 4.829335
## 4 -0.007149860 0.01851852 0.06096056   36798 4.850823
## 5 -0.022536940 0.01851852 0.06057517   27278 3.569431
## 6 -0.004935614 0.01851852 0.06166893   25001 3.223607

The remaining datasets used to support the CNV data display on the BubbleTree plots.

load(system.file("data", "cancer.genes.minus2.rda", package="BubbleTree")) # list of 379 known cancer genes
load(system.file("data", "vol.genes.rda", package="BubbleTree")) # another list of 105 known cancer genes
load(system.file("data", "gene.uni.clean.gr.rda", package="BubbleTree"))
load(system.file("data", "cyto.gr.rda", package="BubbleTree")) # load cytoband coordinate data
load(system.file("data", "centromere.dat.rda", package="BubbleTree")) # load centromere coordinate data
load(system.file("data", "all.somatic.lst.RData", package="BubbleTree")) # load SNV location data
load(system.file("data", "allHetero.lst.RData", package="BubbleTree")) # load sequence variants
load(system.file("data", "allCNV.lst.RData", package="BubbleTree")) # load copy number variation data
load(system.file("data", "hg19.seqinfo.rda", package="BubbleTree")) # load hg19 sequence data

2 Main Bubbletree Functions

2.1 BubbleTree model and diagram

BubbleTree is a model based on three valid assumptions: 1) the paired normal specimen expresses the common diploid state, 2) variant sites are bi-allelic, and 3) genome segments (rather than the whole genome) with homogeneous copy number ratio and BAFs, exist in the profiled tumor genome. The first two assumptions generally hold, whereas the last homogeneity assumption can also be satisfied even in the case of a complex tumor clonal structure.

As the three assumptions are all generally plausible, we therefore developed a model for the BubbleTree diagram. For one homogenous genomic segment (x:y;p), we have,

Expected copy number, (CN)=2×(1-p)+(x+y)×p

Copy Ratio, R=(CN)/2=(1-p)+(x+y)/2×p (1)

B allele frequency, BAF=(y×p+1×(1-p))/((x+y)×p+2×(1-p))

and the homozygous-deviation score (HDS),

HDS= ∣BAF-0.5∣=(p×∣y-x∣)/(2×[(x+y)×p+2×(1-p)]) (2)

Based on equations (1) and (2), we are able to calculate an R score (copy ratio) and HDS for a segment (x:y; p). For example, (0:1; 0.75) will provide 0.625 and 0.3 for the R scores and HDS, respectively.

2.2 Description of the BubblePlot Graph

BubbleTree plots for Primary Liver Tumor

2.2.1 The Branches

The above plot introduces the relationship between HDS and R score (copy number ratio), both used to elucidate the tumor cell prevalence, ploidy state, and clonality for a tumor sample. Generally, the R score indicates the copy number change, ranging from 0 (homozygous deletion) to 3 (hexaploidy) or higher, while the HDS represents LOH, ranging from values of 0 to 0.5 (i.e., LOH with 100% prevalence). Each branch in the diagram starts at the root (1,0), a value of HDS=0 and R score=1. Namely, a diploid heterozygous genotype segment has a copy number ratio, or R score of 1 (tumor DNA copies=2; normal DNA copies=2, so 2/2=1) with no LOH (HDS=0) and is indicated with a genotype of AB, where the A allele is from one parent and the B allele is from the other parent presumably. Then from the root (1,0), the segment prevalence values are provided in increasing increments of 10%, with each branch representing a different ploidy state. As the values increase along the y-axis, the occurrence of LOH increases, such that on the AA/BB branch at HDS=0.5 and R score=1, this indicates a disomy state with LOH and 100% prevalence for the segment.

Generally, the branches mark the projected positions of segments at the given integer copy number ploidy states and prevalence. The plot clearly highlights how high prevalence values create distinct separation between branches (i.e., ploidy states), while as prevalence approaches zero, the branches are non-distinguishable. The ploidy states of Φ, AABB, and AAABBB all have HDS scores of 0, which indicate no LOH at increasing or decreasing R scores from a value of 1, and therefore differ most from the copy number neutral heterozygous disomy state (AB) by R score only. These three ploidy states indicate homozygous deletion (Φ) or amplifications (AABB=1 DNA copy number gain each allele, AAABBB=2 DNA copy number gains each allele). Other ploidy states such as ABB (brown), ABBB (blue), ABBBB (green), or ABBBBB (purple) share a piece of the same branch (i.e., the indistinguishable branches), suggesting the existence of multiple likely combinations of prevalence and ploidy states for that region. A tumor clone usually has more than one SCNA, so the abundance of the clone can still be inferred from other distinguishable branches.

2.2.2 The Bubbles

Along with the branches from the prediction of the model, bubbles (i.e., the leaves) are depicted on the basis of the real data, where the size of the bubbles are proportional to the length of the homogenous segments. A bubble (i.e. the homogeneous SCNA segment) represents the HDS and R score as measured from the assay, such as WES or WGS data. The location of the bubble determines the allele copy number(s) and prevalence for the SCNA segment. A close proximity between a bubble and branch indicates an integer copy-number (e.g. 15q11.2-14), whereas any deviation between the bubble and branch (e.g, 7q21.11-21.12) is due to either variation in the measurement or a non-integer copy-number – something that occurs with multiple clones harboring different SCNAs over the same region.)

2.3 This will create BubbleTree plots for all 41 samples using adjusted and non-adjusted CNV data.

library(BubbleTree)
load(system.file("data", "allCall.lst.RData", package="BubbleTree"))

btreeplotter <- new("BTreePlotter", max.ploidy=5, max.size=10)

pdf.filename <- paste("adjplot_all", "pdf", sep=".")
pdf(pdf.filename, width=6, height=4)

l_ply(names(allCall.lst), function(nn){
    cat("Processing sample name: ", nn, "\n")
    rbd1 <- allCall.lst[[nn]]@rbd
    rbd2 <- allCall.lst[[nn]]@rbd.adj
    arrows <- trackBTree(btreeplotter, rbd1, rbd2, min.srcSize=0.01, min.trtSize=0.01)
    btree <- drawBTree(btreeplotter, rbd1) + drawBubbles(btreeplotter, rbd2, "gray80") + arrows
    print(btree)
})
dev.off()
Adjusted and Non-Adjusted CNV BubbleTree Plots for Primary Liver Tumor

Adjusted and Non-Adjusted CNV BubbleTree Plots for Recurrent Liver Tumor

2.4 This will create BubbleTree plots for all 41 samples using only non-adjusted CNV data.

library(BubbleTree)
load(system.file("data", "allCall.lst.RData", package="BubbleTree"))

btreeplotter <- new("BTreePlotter", max.ploidy=5, max.size=10)

pdf.filename <- paste("adjplot_single", "pdf", sep=".")
pdf(pdf.filename, width=8, height=6)

l_ply(names(allCall.lst), function(nn){
    cat("Processing sample name: ", nn, "\n")
    rbd1 <- allCall.lst[[nn]]@rbd
    rbd2 <- allCall.lst[[nn]]@rbd.adj
    arrows <- trackBTree(btreeplotter, rbd1, rbd2, min.srcSize=0.1, min.trtSize=0.1)
    btree <- drawBTree(btreeplotter, rbd1)  + arrows
    print(btree)
})
dev.off()
Non-Adjusted CNV BubbleTree Plots for Primary Liver Tumor

Non-Adjusted CNV BubbleTree Plots for Recurrent Liver Tumor

2.5 To generate a Excel report of a BubbleTree analysis. This report only shows samples that have tumors with high ploidy and high purity.

library(BubbleTree)
load(system.file("data", "allRBD.lst.RData", package="BubbleTree"))
btreepredictor <- new("BTreePredictor")
btreepredictor@config$cutree.h <- 0.15

high.ploidy <- rep(TRUE, length(allRBD.lst))
high.purity <- rep(TRUE, length(allRBD.lst))

high.ploidy[c("sam6",
              "ovary.wgs",
              "ovary.wes",
              "TCGA-06-0145-01A-01W-0224-08",
              "TCGA-13-1500-01A-01D-0472-01",
              "TCGA-AO-A0JJ-01A-11W-A071-09")] <- FALSE

high.purity[c("sam6", "ovary.wgs", "ovary.wes")] <- FALSE


allCall.lst <- plyr::llply(names(allRBD.lst), function(nn) {
    cat("Processing sample name: ", nn, "\n")
    rbd <- allRBD.lst[[nn]]
    btreepredictor@config$high.ploidy <- high.ploidy[nn]
    btreepredictor@config$high.purity <- high.purity[nn]
    btreepredictor <- loadRBD(btreepredictor, rbd)
    btreepredictor@config$min.segSize <- ifelse(max(btreepredictor@rbd$seg.size, na.rm=TRUE) < 0.4, 0.1, 0.4)
    
    btreepredictor <- btpredict(btreepredictor)
    
    cat(info(btreepredictor), "\n")
    return(btreepredictor)
})

names(allCall.lst) <- names(allRBD.lst)
results <- list()
for (name in names(allCall.lst)) {
    results[[name]] <- allCall.lst[[name]]@result$dist
}

xls.filename <- paste("all_calls_report", "xlsx", sep=".")
print(xls.filename)
saveXLS(results, xls.filename)

2.6 To perform a BubbleTree analysis with an overlay of 77 common cancer genes.

library(BubbleTree)
load(system.file("data", "allCall.lst.RData", package="BubbleTree"))
load(system.file("data", "cancer.genes.minus2.rda", package="BubbleTree"))
load(system.file("data", "vol.genes.rda", package="BubbleTree"))
load(system.file("data", "gene.uni.clean.gr.rda", package="BubbleTree"))
load(system.file("data", "cyto.gr.rda", package="BubbleTree"))

# 77 common cancer genes
comm <- btcompare(vol.genes, cancer.genes.minus2)

btreeplotter <- new("BTreePlotter", branch.col="gray50")
annotator <-new("Annotate")

pdf.filename <- paste("bubbletree_preview", "pdf", sep=".")
pdf(pdf.filename, width=10, height=6)

allPreview <- llply(names(allCall.lst), function(nn){
    cat("Processing sample name: ", nn, "\n")
    cc <- allCall.lst[[nn]]
    z <- drawBTree(btreeplotter, cc@rbd.adj) + ggplot2::labs(title=sprintf("%s (%s)", nn, info(cc)))
    print(z)
    out <- cc@result$dist  %>% filter(seg.size >= 0.1 ) %>% arrange(gtools::mixedorder(as.character(seqnames)), start)
    
    ann <- with(out, {
        annoByGenesAndCyto(annotator,
                           as.character(out$seqnames),
                           as.numeric(out$start),
                           as.numeric(out$end),
                           comm$comm,
                           gene.uni.clean.gr=gene.uni.clean.gr,
                           cyto.gr=cyto.gr)
    })
    
    out$cyto <- ann$cyto
    out$genes <- ann$ann
    return(out)
})
dev.off()
BubbleTree plots for Primary Liver Tumor

BubbleTree plots for Recurrent Liver Tumor

2.7 BubbleTree can create a summary visualization that displays the concordance between copy number and max B-allele Frequency for each chromosome as well as compare the BAF and R scores.

library(BubbleTree)
load(system.file("data", "allCall.lst.RData", package="BubbleTree"))
load(system.file("data", "centromere.dat.rda", package="BubbleTree"))
load(system.file("data", "all.somatic.lst.RData", package="BubbleTree"))
load(system.file("data", "allHetero.lst.RData", package="BubbleTree"))
load(system.file("data", "allCNV.lst.RData", package="BubbleTree"))
load(system.file("data", "hg19.seqinfo.rda", package="BubbleTree"))

trackplotter <- new("TrackPlotter")

gr2 = centromere.dat

pdf.filename <- paste("all_tracks3", "pdf", sep=".")
pdf(pdf.filename, width=10, height=7)

plyr::l_ply(names(allCall.lst), function(nn) {
    cat("Processing sample name: ", nn, "\n")
    ymax <- ifelse(nn %in% c("lung.wgs", "lung.wes"), 9, 4.3)
    
    p1 <- xyTrack(trackplotter,
                  result.dat=allCall.lst[[nn]]@result,
                  gr2=gr2,
                  ymax=ymax) + ggplot2::labs(title=nn)
    
    p2 <- bafTrack(trackplotter,
                   result.dat=allCall.lst[[nn]]@result,
                   gr2=gr2,
                   somatic.gr=all.somatic.lst[[nn]])
    
    t1 <- getTracks(p1, p2)
    
    z1 <- heteroLociTrack(trackplotter, allCall.lst[[nn]]@result, gr2, allHetero.lst[[nn]])
    z2 <- RscoreTrack(trackplotter, allCall.lst[[nn]]@result, gr2, allCNV.lst[[nn]])
    t2 <- getTracks(z1, z2)
    
    gridExtra::grid.arrange(t1,t2, ncol=1)
    
})
dev.off()
CNV and BAF Plots for Primary Liver Tumor

CNV and BAF Plots for Recurrent Liver Tumor

2.8 To perform a comparison of cancer datasets.

library(BubbleTree)
load(system.file("data", "allCall.lst.RData", package="BubbleTree"))

btp <- new("BTreePlotter", max.ploidy=5, max.size=10)
pairs <- list(c("lung.wes", "lung.wgs"),
              c("HCC4.Primary.Tumor", "HCC4.Recurrent.Tumor"),
              c("HCC11.Primary.Tumor", "HCC11.Recurrent.Tumor" ))

pdf("medi.comparePlot.single2.pdf", width=8,height=6)

plyr::l_ply(pairs, function(p){
    cat(p, "\n")
    rbd1 <- allCall.lst[[p[1]]]@result$dist
    rbd2 <- allCall.lst[[p[2]]]@result$dist
    
    srcSize <- 0.5
    trtSize <- ifelse(p[1] == "lung.wes", 0.1, 1)
    minOver <- ifelse(p[1] == "lung.wes", 5e6, 1e7)
    arrows <- trackBTree(btp,
                         rbd1,
                         rbd2,
                         min.srcSize=srcSize,
                         min.trtSize=trtSize,
                         min.overlap=minOver)
    
    z <- drawBTree(btp, rbd1)
    if(!is.null(arrows))
        z <- z + arrows + ggplot2::labs(title=sprintf("%s -> %s", p[1], p[2]))
    print(z)
})
dev.off()
Comparison of HCC11 Primary and Recurrent Tumors

2.9 To print out an Excel document of summary of the pre-called CNV data.

library(BubbleTree)
load(system.file("data", "allCall.lst.RData", package="BubbleTree"))

all.summary <- plyr::ldply(allCall.lst, function(.Object) {
    purity <- .Object@result$prev[1]
    adj <- .Object@result$ploidy.adj["adj"]
    ploidy <- (2*adj -2)/purity + 2  # when purity is low the calculation result is not reliable
    
    with(.Object@result,
         return(c(Purity=round(purity,3),
                  Prevalences=paste(round(prev,3), collapse=", "),
                  "Tumor ploidy"=round(ploidy,1))))
}) %>% plyr::rename(c(".id"="Sample"))

xls.filename <- paste("all_summary", "xlsx", sep=".")
saveXLS(list(Summary=all.summary), xls.filename)